Description Usage Arguments Value
ensemble
1 2 3 4 |
mlout |
Inputs an mlblob object containing multiple models from the result of the startml function |
algorithm |
Character. Algorithm used to create the ensemble. Can be one of the algorimths from startml, or "mean," "median," or "vote." Default is mean. |
validation |
Character. Type of validation to be used in ensemble, can be "random_holdout" or "xval" for cross validation. Default is xval. |
percent_reduce |
Numeric. Indicates the rough percentage to reduce input dementions by. Default is NULL. |
reduce_method |
Character. Indicates the dimentional reduction method to use. Can be "PCA" or "auto_encoder." When "auto_encoder" algorithm must be deeplearning, this can greatly increase the length of time nessesary to run an ensemble, but may result in a more stable model. Default is NULL. |
keep_features |
Boolean. When TRUE, orginial input features are mixed with weak learner predictions to create the new input data set for the ensemble. When FALSE, only weak learner esimates are used. Default is FALSE. |
grid_search |
Boolean. Use a grid search to find optimal hyper parameter when algorithm is "gbm" randomForest" or "deepleaning." Default is FALSE, this can greatly increase the ensemble training time but may result in a superior parameter. |
runtime_secs |
Numeric. Number of seconds to run grid search when grid_search is TRUE. Default is 60 seconds. |
wd |
directory to save ensemble. Default is current directory |
mlblob object including ensemble model, ensemble input data set, and predictions on train, valid, test, and new_data sets if available.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.